Schemes and versioning

In section 4.3 of the 8 January 2001 XPointer Working Draft 
(http://www.w3.org/TR/xptr#schemes), the following appears:
>Each XPtrPart begins with a Scheme that identifies the particular notation 
>used for that XPtrPart. This specification defines only two schemes, 
>xpointer and xmlns, and reserves all others when the media type of the 
>referenced resource is one of text/xml, application/xml, 
>text/xml-external-parsed-entity, or 
>application/xml-external-parsed-entity. However, the scheme mechanism 
>provides a general framework for extensibility that can be used for future 
>versions of XPointer, or for other media types that wish to adopt all or 
>part of this specification in defining their own fragment identifier 
>languages.

I'm concerned that developers with needs other than the complete set of 
XPointer functionality are poorly served by the current single and 
unversioned xpointer scheme.

It seems like an xpath scheme - which only contained XPaths and not the 
extensions XPointer provides - would allow developers to use XLink for 
node-to-node hypertext applications at much lower cost than understanding 
and implementing the full range of possibilities in XPointer.

Applications which fully understood XPointer would also understand the 
XPath scheme, but they could provide additional 
functionality.  Applications which only understood XPath would be able to 
point to the xpointer() content of the URI and explain that they didn't 
understand it.

This approach would allow more considerably more functionality than is 
required (http://www.w3.org/TR/xptr#app-conformance) by minimal 
conformance, while still not imposing the larger costs of full conformance.

It seems like schemes are a useful tool for encouraging diversity, but the 
XPointer specification in its current form does not appear to apply it that 
way.

If it were possible - and I'm aware certain people believe changes neither 
possible nor valuable - it seems like it might be reasonable to publish one 
document creating an XPointer framework.  This would contain the material 
required by minimal conformance, possibly child sequences, the rules for 
the xmlns scheme, and rules for creating other schemes.  That could be 
published immediately as XPointer 1.0.

The details of what goes in XPointer could then be reserved for 
explanations of the schemes, and it seems reasonable to suggest that 
multiple schemes are possible and probably even to be encouraged.


Simon St.Laurent - Associate Editor, O'Reilly & Associates
XML Elements of Style / XML: A Primer, 2nd Ed.
XHTML: Migrating Toward XML
http://www.simonstl.com - XML essays and books

Received on Tuesday, 8 May 2001 09:23:14 UTC